CitedEvidence
CitedEvidence combines paper discovery with document reading and academic writing tools. Its public research playground accepts a question, a Boolean search string or an entity lookup. Researchers can then move into modules for reading PDFs, collecting evidence and drafting academic work. The publisher also offers Scholar API and MCP access for applications that need research records.
Start with a question and inspect the records
The discovery interface covers works, authors, journals, institutions, funders, publishers and topics. Visible filters include open access and recent publication. CitedEvidence lists research sources including PubMed, Semantic Scholar, Crossref and arXiv. Those source names help explain the intended discovery workflow, but they do not establish complete coverage of a field.
The publisher advertises more than 300 million metadata records and 50 million file-access records. It says file formats vary by paper. Treat those totals as vendor figures, then inspect the particular record you intend to use: title, authors, publication date and access status are more useful than a catalogue-wide count when locating one study.
For a literature search, keep the question and the resulting search string together. Review whether the returned papers address the population or method you need before asking a writing module to synthesize them. The public page describes an AI snapshot when its public model is configured; availability of that feature should be checked in the current interface.
Read the paper before carrying a claim into a draft
The PDF Chat module accepts uploaded papers and describes targeted questions about methods, findings and limitations. It aims to keep answers connected to the document. That is a useful reading workflow, provided you return to the cited passage to check the answer. A concise generated summary can leave out conditions that matter to the result.
CitedEvidence's Deep Reasoning module describes planning research tasks, collecting evidence and producing structured academic outputs. The wider suite includes citation management, manuscript review and thesis-to-manuscript conversion. These are separate tasks: discovering a relevant paper does not verify that a later paragraph represents it correctly.
Check access and subscription details for the module you need
The site allows exploration without logging in, while its feature pages describe account and subscription paths into particular modules. The inspected pages do not establish one universal price or unrestricted access to the full suite. Confirm the required module and its current allowance before uploading a large reading set.
For builders, Scholar exposes metadata, available files and provenance fields such as source database, record ID and access status. Preserve those fields in your own application so readers can trace an answer back to the record. Keep access permissions separate from discovery: finding a PDF link does not establish a right to redistribute the paper.